Every term the site throws at you, defined once, in one place. For a guided tour of the card layout itself, see How to Read a Card; for the full derivation of the rating, see Net Goals 101. All ratings are model estimates, not official NHL statistics.
Core units
Net goals added
The site’s common currency: how many goals a player adds (or costs) his team versus a league-average player, in calibrated model units. It’s usually quoted per 84 games, a full season, so a 300-game career and a 1,000-game career sit on the same ruler. The team-page forecast tables show season totals instead (per-84 rate × projected games).
xG (expected goals)
The probability that a given shot becomes a goal, based on its location, type, and game situation (strength state, rebound within 3 seconds, rush chances). Summing xG over shots gives expected goals for/against. In other words: a rebound at the top of the crease and a wrister from the point aren’t the same shot, and xG doesn’t pretend they are.
WPA (win probability added)
A weight on each chance for how much it moves the odds of winning. A chance in a tied third period and the same chance up four are not the same chance, and everyone in the building knows it. WPA puts a number on the difference, and that’s how expected-goal flow gets priced into wins.
rel (relative to context)
Measured against what a league-average player would produce in the same spot: score state, home/road, zone starts, strength state. Since 2026-07 the components are also re-centered so the league average is exactly zero in every season.
rel-xG-WPA Elo
The engine under the on-ice ratings: win-probability-priced expected-goal flow, measured relative to league-average context, run through a game-by-game Elo update.
Rating components
Overall
The sum of the ten skater components below. For goalies, Overall is GSAx per 84 games instead (the skater components don’t apply), net of the part his defenders’ Suppression explains, plus the mirror of the shootout attempts he faced.
EV Offense / EV Defense
On-ice expected-goal creation (offense) and suppression (defense) at even strength versus league-average context, scaled to the calibrated marginal value. Five skaters share the on-ice flow, so the marginal is ~4.4× smaller than face value: nobody gets full credit for a goal four other guys were on the ice for. EV Offense is net of Playmaking: the share of linemates’ chances a player created is moved to the creator; EV Defense is net of Suppression the same way.
PP / PK
The same priced on-ice flow on the power play and penalty kill.
Finishing
Goals above expected (GAx) on the shooter’s own shots on goal: did he beat the goalie more often than those looks deserved? Individual credit, at face value, with one exception, regular-season overtime: an OT goal ends the game and moves the scoring team from about 1.5 to 2 expected points, so it counts as about 1.5 regulation goals (less on an OT power play, whose team was already favoured). The shooter keeps half; the other half of each shot’s goals above expected goes to his on-ice teammates’ Playmaking, because the guy who put it on his tape earned a cut.
Playmaking
How much the player raises his linemates’ offense: the pass before the goal, priced. It comes in two parts. Shot quality: expected goals per unblocked attempt with him on the ice, from joint regressions over every attempt since 2010-11 that hold each shooter’s own skill fixed and account for the teammates and opponents on the ice. Assists: half of every shot’s goals above expected is shared among the shooter’s on-ice teammates; on a goal the primary assist gets 4 shares, the secondary 2 and every other teammate 1, while the expected-goal cost of every shot is split equally. The credit is moved, not added: it comes out of the on-ice rows every skater shares (EV Offense, PP) and out of the shooter’s Finishing, so no team’s total changes. Playoff values use the regular-season estimates.
Suppression
The defensive mirror of Playmaking: how much the player lowers the opponents’ shot quality (expected goals per unblocked attempt) and their finishing (goals against above expected) while he is on the ice, from the same joint regressions with the opposing shooter’s skill and the defending goalie held fixed. Moved, not added: the quality part comes out of the shared EV Defense / PK rows, the finishing part out of the goalie’s GSAx (every goalie value on the site is GSAx net of it), so no team’s total changes. The finishing part is small: once shot quality is priced, skaters barely move save percentage.
Penalties
Penalties drawn minus taken, priced at that season’s measured value of a drawn minor (≈0.16–0.20 goals: the average net goals scored in the two minutes after a penalty is drawn, measured league-wide per season), re-centered so average penalty behavior is zero. Put plainly, every trip to the box you draw (or take) is worth about a sixth to a fifth of a goal.
Faceoffs
Net faceoff wins × 0.010 goals (fixed-effects calibration). Shown for context: post-draw flow already appears in the on-ice components, so this partially overlaps them. A hundred extra draws won is about one goal.
Shootout
The standings value of a skater’s shootout attempts: each attempt is worth the win probability it added given the round, score and who shoots next (a sudden-death winner counts more than a first-round goal), at 3 goals per shootout win (6 goals per win). The goalie he faced loses the same amount, inside his GSAx. Real points, but shootout success barely repeats (season-to-season correlation ~0.04), so career values are shrunk by career attempts (half signal at about 86), like Finishing by shots. Regular seasons 2010-11 on; no playoff shootouts.
Goaltending
GSAx (goals saved above expected)
Expected minus actual goals against on the shots faced, each shot weighted by what conceding it would cost his team in win probability at that score and time (a goal against while protecting a late one-goal lead counts far more than one down three; the average shot counts 1). Positive means the goalie stopped more than the shots he faced predicted. Because of the weighting it isn’t exactly xGA − GA.
xSv%
The save percentage an average goalie would post on the same shots; compare with actual Sv%.
Reb/100 vs lg
Rebounds yielded per 100 shots faced, minus the league rate that season (0 = average; lower is better) — rebound control is where goalie skill persists most. Season-relative because the raw league rate roughly doubled between 2019 and 2025 through play-by-play recording changes, not because goalies forgot how to smother pucks.
Card panels
NHL EDGE (2021+)
Puck-and-player tracking from the NHL’s EDGE system, which only exists from 2021-22 on (longer careers start blank): hardest shot and top skating speed in mph, and 20+ mph speed bursts per game. Boxes are colored by where the number ranks within that season’s tracked cohort (skaters with ≥50 shots on goal); the career column is the career best (bursts: the career rate).
Competition / Teammates (on-ice context)
Who a player skated against and with, season by season: the shared-TOI-weighted average Overall net goals/84 of the opposing skaters he faced (Competition) and of his own linemates (Teammates), goalies excluded on both sides. The panel also lists each season’s four most common teammates with the share of the player’s ice time they were on for.
84-game form
The career’s shape in one line: a trailing 84-game moving average of the Overall rating, in the same net-goals-per-84 units. The dot marks the career peak and the number at the right end is the current value. Even elite careers dip below zero here: 84 games is still a small sample, as anyone who’s paid for a contract year knows. Goalie cards draw the same line for GSAx/84 over the last 84 appearances.
Career net goals (cumulative)
The running career total of the same per-game net goals the 84-game form line averages: how much value the career has banked versus a league-average player, in goals. A flat stretch is league-average play; a rising slope is value being added. Goalie cards accumulate GSAx instead.
Shot mix — type & location
The career shot diet: bars show the share of the player’s shots on goal by type, each priced as net goals per 84 (goals above xG on those shots; types with fewer than 100 career shots are unpriced). The map colors each zone of the offensive half by his share of shots from that spot versus the league average for his position — blue means more than a typical F or D shoots from there, red less.
Save map
The goalie equivalent: each zone colored by save percentage above or below what that zone’s shot locations predict.
Projections & simulation
Projected value
The simulation’s forward-looking estimate for 2026-27: recent seasons weighted by a 0.55 decay, age-translated using the skill aging curves, and shrunk toward the mean for small samples. The Projected columns on the site combine its three parts: on-ice value + finishing GAx + net penalties.
Projected GP / expected games
Each rostered player’s expected games played, from durability history. Forecast tables multiply per-84 rates by this to get season totals.
Projected standings
10,000 Monte Carlo simulations of the season: 12F/6D/2G rosters taken from Daily Faceoff’s projected line combinations (every player still priced by the model itself), team strength from the projected player values, game outcomes from a calibrated win-probability model including back-to-back rest edges.
Playoff %
The share of the 10,000 simulated seasons in which the team makes the playoffs.
Δ (rank change)
Projected league-wide finish compared to the team’s actual 2025-26 finish.
Small-sample shrinkage
Career rates in team tables are regressed toward zero by GP/(GP+40), so a hot October doesn’t read like a career.
Player archetypes
How archetypes are built
A k-means clustering of skaters (100+ GP, since 2010-11, goalies excluded) over 25 style features: the nine rating components (Playmaking and Suppression among them), EV/PP/SH ice time, goals, assists, PIM, shots, hits, blocks, takeaways/giveaways, finishing per shot, slap/backhand/tip shot shares, and rebound generation (faceoffs are dropped for defensemen, who almost never take one). Every feature is measured relative to the player’s own era (z-scored within each season and position group), averaged over the player’s last 100 GP — so the type describes who he is now, not a career blend: a veteran whose minutes and results have faded is typed by his current role. Each season-end window is typed too, which gives every player an archetype history. A 2012 grinder and a 2025 grinder land in the same bucket. Forwards form 8 clusters, defensemen 7; the names are editorial, the memberships are the data’s. One forward type is a rule rather than a cluster: Playmaker (below), because clustering never separates the elite creators from the other stars. The two top-pair defense names also promise quality, so they require at least 20 minutes a game and net goals per 84 above replacement level over the window (shrunk for sample size); a defenseman who misses either takes his next-closest type.
First-line star
The franchise offense drivers: elite even-strength and power-play value, top takeaway rates, draw far more penalties than they take. The stars who are not Playmakers — shooters and all-rounders.
Playmaker
The elite creators: a Playmaking component (last 100 GP) at least two standard deviations above other forwards, and worth at least as many goals per 84 games as their own Finishing — they add more through their linemates’ chances than through their own shot. Elite playmakers who add even more as shooters (Kucherov, Pastrnak, Panarin) stay First-line stars. Assigned by that rule after clustering (they would otherwise sit among the First-line stars or top-six snipers).
Top-six sniper
Finishing first: the best goals-above-expected per shot, heavy power-play usage, slap-happy shot diets — and the worst even-strength defense of any forward group.
Two-way top-six forward
Scoring forwards who also do the dirty work: faceoffs, penalty kill, net-front tips and rebound generation alongside above-average offense.
Top-six play-driver
Top-six even-strength and power-play minutes and positive even-strength offense, but below-average finishing — they move play forward more than they convert it.
Complementary skill winger
Skill without the heavy lifting: decent finishing in softer minutes, little shorthanded, blocking or physical involvement.
Two-way PK forward
Middle-six forwards who kill a lot of penalties and still play real even-strength minutes: shorthanded time, blocks and shot suppression, with some scoring.
Checking-line PK forward
Bottom-six penalty killers: the most shorthanded time among forwards with the least even-strength and power-play time; blocks, hits, little scoring.
Physical enforcer
Penalty minutes more than two standard deviations above average, big hit totals, the fewest minutes — and they take far more penalties than they draw.
Franchise No. 1 defenseman
The elite cluster: the model’s best even-strength and power-play value, takeaways and minutes (23–24 a game) — Makar, Hughes, Fox, Hedman, Josi, Werenski. Needs 20+ minutes and above-replacement net goals.
Top-pair puck-mover
Big minutes (about 22 a game) with power-play time and playmaking, but not the finishing — Slavin, Heiskanen, Ekholm, Toews. Needs 20+ minutes and above-replacement net goals; big-minute puck-movers below replacement are typed by their next-closest style.
Shooting defenseman
Big minutes with the best finishing and goals per 84 of any defense group — point shots that go in (Burns, Weber, Faulk, Morrissey).
Stay-at-home defenseman
Modest minutes, little power play, a net-front shot diet of tips and backhands — the stay-at-home depth pairing (Vlasic, Dumoulin, Oleksiak).
Shot-blocking PK defenseman
Defense-first: heavy shorthanded time and shot-blocking, about 20 minutes, essentially no power play.
Physical shutdown D
The heavy games: hits and penalty minutes well above average in modest minutes, takes more penalties than he draws.
Sheltered third-pair D
The fewest minutes at even strength and shorthanded, no penalty kill; never used for a defenseman playing 20+ minutes.
Steadiness
How steadiness is measured
The card header’s steadiness tag describes how a skater’s value is spread across games, not how much of it there is. For every game we take the player’s net goals and measure how deep the below-average games go (the root-mean-square of the negative games), then compare that with skaters of the same position, net goals per 84 and ice time. The career score is games-weighted and shrunk toward average for short careers (half weight at about 60 GP for forwards, 85 for defensemen), and the percentile ranks active skaters at the position with 100+ career GP (higher = steadier). Split between odd and even seasons, careers agree closely (r 0.83 forwards, 0.75 defensemen), so it’s a real, repeatable trait.
What it is not
Not a value rating: at equal net goals/84 a boom-or-bust player is worth exactly as much as a steady one. Not hot and cold streaks either: which players run hot or cold for weeks at a time doesn’t carry over from one season to the next, so the model doesn’t rate it. Streaks happen; streaky players, as a type, don’t show up. Shooters skew boom-or-bust because goals come in lumps; playmakers and shutdown defensemen skew steady.
Very steady
Top 10% at the position.
Steady
70th to 89th percentile.
Average
31st to 69th percentile (shown on cards as “Average steadiness”).
Boom-or-bust
11th to 30th percentile.
Very boom-or-bust
Bottom 10% at the position.
Data conventions
Career weighted average
Career values are games-played-weighted means over seasons, not simple averages, so a 20-game season counts less than a full one.
Empirical-Bayes shrinkage
Small samples are pulled toward the league average before being called skill: on the cards and the Players page, each component’s career rate is blended with the league mean using a prior worth a component-specific number of games, so 40 hot games move a career number far less than 400 do. Per-season values are never shrunk.
Active players
The Players table includes everyone on a 2026-27 cap sheet who either played in 2025-26 or is on a current NHL roster — this keeps injured players and call-ups while excluding retired dead-cap contracts (buyouts, LTIR) — plus unsigned free agents: restricted free agents under their 2025-26 team with “RFA” in the cap column (assumed re-signed), unrestricted free agents under team “UFA” with their own card deck.
Cap hit
2026-27 cap hit from the cap-sheet snapshot; contract efficiency on the cards: player value (what the league pays, as a share of the cap, for a player’s 3-season net goals/84) minus the cap hit.